If I have a solved Rubik's cube, and I make a finite sequence of (legal) moves repeatedly, prove that I will eventually resolve the puzzle.

(this wording is kinda goofy but i hope its sorta intuitive)

if u say "eventually," doesn't that mean an infinite number of moves?
I mean all rubik's cubes can be solved from any position in 20 moves, so with luck and infinite time you will get those 20 or less moves all correct.

oh i've thought about this before
imagine all possible rubik's cube positions as vertices, and put them on a directed graph, with an edge from one position to another if the sequence turns the first position into the second
note that every vertex has an indegree of 1 and outdegree of 1
now assume you can't resolve the puzzle, that means you get stuck in a cycle not containing the original position
but to get into such a cycle without the original position, there must be a vertex with an indegree greater than 1 (not sure how to rigorously show this, but inuitively makes sense), which is a contradiction

By the Pigeonhole Principle, there must be some state of the cube that is achieved twice after applying this algorithm arbitrarily many times. Therefore, repeating the algorithm some number of times brings a state back to itself, and hence it will bring the solved state back to itself.

dang well said
that's smart

Theorem 4.2.2 provides an upper limit with only face turns (standard HTM) of moves.

I must be missing something, but what if(theoretically) every position except the solved state is in a loop? Yes, the first unsolved position will repeat, but that yields no insight on why the solved face should repeat?

In theory, that solution would take a long time, but it will happen, right? Imagine it like a circle. Every time you bend the circle to make another, the start still remains, even though there is a detour. In a sense, it would loop back, after many loops. (I have no idea btw)

The generalization of this is phrased in terms of group theory: "in a finite group, every element has a finite order". Here the group is the permutations you can make on a rubiks cube through legal moves. the order of a certain combination of moves is how many times it needs to be repeated to go back to the "nothing happened" outcome (called the "group identity").

Wait is this true? This only implies that some state might repeat, doesn't mean that the initial state will.

correct me if im wrong but if its not the initial state then its not a group (im rusty at group theory tho so)

I'll address these two concerns with my solution. Write the algorithm as . Then suppose and correspond to the same state (some distinct values of and must satisfy this by PHP). Therefore, applying the algorithm times to a state returns that state back to itself. This is true no matter what state we are currently at, since every piece is unique on the cube. This means that after algorithms at the beginning, we will return to the solved state.

no i mean i have a certain sequence of moves, maybe like ULU that i repeat until it solves itself

oop im not familiar with rubik's cube rules, but "legal" moves don't incorporate like turning a side back and forth right; cuz that's like a finite sequence?

If you turn a side back and forth by and then then trivially the puzzle gets resolved after applying this finite sequence of legal moves.

when i was small i counted all 1260 moves and surprisingly didn’t mess up (so much dedication)

:skull: mb

some clarifications:

no i mean like no breaking the cube, only twisting the sides, back and forth is ok.

yes, however the problem is to prove that every sequence eventually solves the cube, not to find just one.

I was responding to #16 not the original post.